基于电容原理的谷类作物茎蘖数测量技术研究

吴董雪; 谷慧杰; 李保儒; 唐晓灿; 刘秀位

doi:10.12357/cjea.20250183

基于电容原理的谷类作物茎蘖数测量技术研究

Research on stem/tiller number measurement technology of cereal crops based on capacitance principle

摘要

摘要: 茎蘖数是谷类作物产量的重要构成因素, 高通量监测作物密度对品种选育和栽培管理至关重要。然而, 传统人工计数方法效率低下, 现有自动化测量方法存在高密度场景测量精度不足、设备成本高及田间动态监测能力受限等瓶颈。因此, 亟待开发高效且低成本的茎蘖数测量技术。本研究提出基于电容原理的茎蘖数测量新技术, 以谷类作物小麦和燕麦为对象, 通过理论建模、室内茎秆模拟、盆栽密度试验、田间品种与密度结合试验及茎数估测模型构建系统验证其可行性。理论模型构建表明, 茎蘖数与电容值的线性关系源于介电常数差异与电场调制效应。Maxwell-Garnett理论及电路模型均表明, 茎蘖数与电容变化量呈线性正相关。通过对不同试验条件下(电极配置、测量配置等)的测量结果进行分析, 发现在采用直径6 mm、间距3 cm、长度30 cm的电极棒且随机接触测量时, 电容与茎蘖数线性关系最好(R²≥0.92, RMSE≤29 茎∙m⁻², P<0.01)。田间试验表明, 数字电桥结合四端电桥法、采用平行电极以及配置接地电极棒的测量配置时, 茎蘖数与电容间的线性相关性最好(R²=0.69, P<0.01, RMSE=79茎∙m⁻²)。此外, 不同生育期的测量结果表明, 两种作物的茎蘖数与电容之间均呈线性相关, 尤其在茎蘖旺盛时期, 电容对茎蘖数的响应也达到显著水平(小麦分蘖期: R²=0.46, RMSE=103茎∙m⁻², MAE=80茎∙m⁻², P<0.01; 燕麦分蘖期: R²≥0.55, RMSE≤149茎∙m⁻², MAE≤117茎∙m⁻², P<0.01)。在建立田间茎蘖数估测模型时, 采用支持随机森林回归模型可精准测量小麦和燕麦的茎蘖数(小麦: R²=0.76, RMSE=160茎∙m⁻², MAE=123茎∙m⁻², P<0.05, n=102; 燕麦: R²=0.74, RMSE=110茎∙m⁻², MAE=89茎∙m⁻², P<0.05, n=143)。本研究结果充分表明, 基于电容原理可以有效地测量田间谷类作物的茎蘖数, 特别是茎蘖数与电容之间的稳定线性关系为研发低成本且高效的茎数测量技术提供了新思路。

Abstract: Stem/tiller number is a critical component of cereal crop yield, and high-throughput monitoring of crop density is essential for variety selection and cultivation management. However, traditional manual counting methods are inefficient, while existing automated measurement methods suffer from bottlenecks such as insufficient accuracy in high-density scenarios, high equipment costs, and limited field dynamic monitoring capabilities. Therefore, there is an urgent need to develop efficient and low-cost stem/tiller number measurement technologies. This study proposes a novel stem/tiller number measurement technology based on the capacitance principle, systematically validated using wheat and oats as objects through theoretical modeling, indoor stem simulation, pot density experiments, field variety-density combined trials, and stem/tiller number estimation model construction. Theoretical model construction reveals that the linear relationship between stem/tiller number and capacitance value originates from differences in dielectric constants and electric field modulation effects. Both Maxwell-Garnett theory and circuit models indicate a linear positive correlation between stem/tiller number and capacitance variation. Analysis of measurement results under different experimental conditions (electrode configurations, measurement setups, etc.) shows that the best linear relationship between capacitance and stem/tiller number (R²≥0.92, RMSE≤29 stem∙m⁻², P<0.01) was achieved using 6 mm diameter, 3 cm spacing, 30 cm length electrode rods with random contact measurement. Field experiments demonstrated that the measurement setup combining a digital bridge with a four-terminal bridge method, parallel electrodes, and a grounding electrode rod yielded the highest linear correlation between stem/tiller number and capacitance (R²=0.69, P<0.01, RMSE=79 stem∙m⁻²). Additionally, measurements across different growth stages showed linear correlations between stem/tiller number and capacitance for both crops, particularly during vigorous tillering stages, where capacitance responses to stem/tiller number were statistically significant (wheat tillering stage: R²=0.46, RMSE=103 stem∙m⁻², MAE=80 stem∙m⁻², P<0.01; oat tillering stage: R²≥0.55, RMSE≤149 stem∙m⁻², MAE≤117 stem∙m⁻², P<0.01). When establishing field stem/tiller number estimation models, the random forest regression model accurately measured stem/tiller number for wheat and oats (wheat: R²=0.76, RMSE=160 stem∙m⁻², MAE=123 stem∙m⁻², P<0.05, n=102; oats: R²=0.74, RMSE=110 stem∙m⁻², MAE=89 stem∙m⁻², P<0.05, n=143). The results of this study fully demonstrate that the capacitance principle can effectively measure stem/tiller number of field cereal crops. The stable linear relationship between stem/tiller number and capacitance provides new insights for developing low-cost and efficient stem/tiller number measurement technologies.

HTML全文

参考文献(66)

施引文献

资源附件(0)